A petition launched last week by economists regarding artificial intelligence’s transformative effect on the economy has secured the support of a glittering list of the profession’s heavyweights. The core demand is that policymakers must immediately intervene to protect society from the potential ill effects of AI.
Its apparent popularity does not, however, render it a consensus view, with some regarding it as the latest example of what could be regarded as excessive government intervention.
The petition – posted on the somewhat alarmist URL www.wemustactnow.ai – is composed of only three brief statements.
One, AI may become radically more powerful over the next 10 years. Two, this could drive an unprecedented transformation of the global economy, larger than the Industrial Revolution, but unfolding over a vastly shorter time frame. It could bring risks, including large-scale job displacement, as well as opportunities such as major gains in living standards. And three, economists, policymakers and technology leaders must act now to understand the economics of transformative AI and to build the incentives, guardrails and institutions needed to steer AI in a direction that complements humans and benefits society.
The list of signatories contains more than 10 Nobel Prize winners, and even includes known libertarian economists such as George Mason University’s Tyler Cowen and Microsoft Research’s Glen Weyl. However, it has also induced a typically flippant response from the suitably self-monikered “Grumpy Economist”, the Hoover Institution’s John Cochrane.
Dr Cochrane expressed relief that policymakers did not try to steer the development of previous revolutionary technologies such as the steam engine, and also pointed out that when the US government took a paternalistic approach to nuclear power, it throttled the industry into submission. He went further by drawing attention to numerous straightforward programmes that governments struggle with, such as public infrastructure and basic education, arguing that this should make us very sceptical of their ability to tackle the far more complex problem of how to manage the development of AI.
These sentiments were humorously echoed by Dr Cowen’s departmental colleague, Bryan Caplan, who quipped: “I favour guardrails for the people who build guardrails for awesome new technologies. They’re the ones with a terrible track record of derailing progress.”
-
Tesla's AI Day 2022 livestream introduced the humanoid robot Optimus as it walks on stage in Palo Alto, California. All photos: AFP -
Optimus waves to guests at Tesla's AI Day in Calfornia. -
Humanoid robot Optimus was on stage as Tesla set out its plans for an AI future. -
Tesla chief executive Elon Musk explains how the Optimus humanoid robot project will evolve. -
The event took place in Tesla's office in Palo Alto, California. -
“There's still a lot of work to be done to refine Optimus and prove it,” Mr Musk told guests at the event. -
Elon Musk said Optimus would be an 'extremely capable robot' that Tesla would aim to produce in the millions and perhaps cost less than $20,000 per unit.
This strikes at the heart of a core societal debate that has existed ever since the 18th century, when the emergence of powerful government bureaucracies gave humans a dilemma over what issues should be controlled by the state. The intellectual sparring has featured contributions from luminaries on both sides of the aisle, such as Adam Smith, Karl Marx, John Stuart Mill and John Maynard Keynes. Though the discourse has many subtleties, in the case of AI, it can be filtered into two key points of emphasis.
Those in favour of government oversight look at historical episodes – such as the Great Depression, modern climate change and the 21st-century emergence of social media – and conclude that doing nothing would be a gross dereliction of duty by both society’s thinkers and its state. In each of these examples, the American state allowed private actors to steer the ship broadly unconstrained by government oversight or direction, resulting in serious societal harm.
Proponents of AI-related state intervention mostly understand that governments – though not all of them – regularly fail to deal with simpler issues, and sometimes make things a lot worse. However, they also see dark clouds gathering on the horizon and would prefer to roll the dice than to just passively wait for history to run its course. When libertarians support stronger interventions in the AI domain, they often argue that society should employ the full range of available policy tools – including government intervention – to reduce the risk of a potentially dystopian future.
Yet the view espoused by sceptics is that entrepreneurs have notable successes in organically taming technologies that initially caused harm. They point to the internet, whose early years saw an explosion in spam, fraud, malware and cybercrime; but which was progressively made safer through market-driven innovations, such as antivirus software, spam filters, encryption, reputation systems and secure online payment platforms.
Likewise, automobiles initially brought unprecedented levels of traffic fatalities, before manufacturers competing for customers developed seat belts, crumple zones, airbags and other safety features that were later widely adopted.
Critically, while governments eventually adopted regulations that aligned with these developments, in a number of countries the civil servants overseeing such efforts were collectively slow and tacitly relied on the initiative that the unencumbered entrepreneurs displayed. From this perspective, the risk is not that governments will do too little, but that they will do too much; freezing a rapidly evolving technology into an inferior trajectory before entrepreneurs have the opportunity to discover solutions that few central planners could have anticipated.
The disagreement is, therefore, not over whether AI poses risks – virtually everyone accepts that it does – but over whether society’s best hope lies in the adaptive power of decentralised markets or the foresight of centralised government.
That is a debate that economists have been having for more than two centuries, and AI is unlikely to settle it. Despite this, it is fair to say that economists are demonstrating more appetite than at any other point in the post-Second World War era for an interventionist stance, stemming from the palpable fear they are experiencing as researchers, employees, parents to young children, adult children to older parents and so on.
Whether that fear proves prescient or misplaced, it has already reshaped one of economics’ oldest debates.
